IOR17 Manuscript FINAL2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

CO2 Foam EOR Field Pilot: Pilot Design, Geologic and Reservoir Modeling, Laboratory

Investigations, and Application of a Reservoir Management Workflow

Z. P. Alcorn*, S. Fredriksen*, M. Sharma†, M. Fernø*, and A. Graue*


* University of Bergen

University of Stavanger

Abstract

A CO2 foam field pilot research program has been initiated to test and advance the technology of CO2
foam systems with mobility control to optimize CO2 integrated EOR and CO2 storage. Previous CO2
foam pilot tests have analyzed field scale displacement mechanisms, foam’s effects on gas mobility,
reservoir injectivity, and overall recovery. Past tests have shown variable amounts of success,
establishing the need for a more integrated methodology for advancing CO2 foam technology for
EOR.

This work describes initial design, generation of geologic and dynamic reservoir models, laboratory
investigations, and the application of a reservoir management workflow for a CO2 foam field pilot in
the Permian Basin of west Texas, USA. Application of a reservoir management workflow guides a
systematic approach from data gathering, model generation, and decision making to final
implementation and analysis of the CO2 foam field pilots. Initial pilot design begins with an improved
reservoir characterization, field pilot selection criteria, and laboratory studies. Laboratory work
investigating foam’s behavior at variable pressures found that increased reservoir pressure will result
in more favorable CO2 foam behavior as it will recover oil more effectively, considering the economic
limits on CO2 and surfactant usage.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Introduction
A multidisciplinary team has been established to combine expertise to develop and test CO2 foam
systems with mobility control at laboratory and field pilot scale to optimize CO2 integrated enhanced
oil recovery (EOR) and CO2 storage. The project commenced in January 2015 with plans to be
completed by the end of 2019. This study details the first phase of project development which
includes field evaluation, initial planning, and design of a CO2 foam for mobility control for EOR
field pilot project at Queen Field in the Permian Basin of west Texas.

Queen Field is an underdeveloped, mature oil field originally discovered in the 1940s. Production
predominantly comes from a heterogeneous sequence of sandstones interbedded with carbonates and
evaporites. Pressure in the reservoir was initially 1500 psia (hydrostatic) and behavior was typical of
solution gas drive with rapidly declining pressure and increasing gas-oil ratios. Currently, the field
contains 74 wells and the reservoir has been depressurized. Operational plans to infill drill and
implement a field wide waterflood are ongoing to repressurize the reservoir, provide useful data
relating to the application of CO2 foam injection, and reestablish the economic viability of the field.

Growing concerns regarding greenhouse gas emissions has led to renewed interest in utilizing CO 2 as
an EOR method as part of Carbon Capture, Utilization, and Storage (CCUS). The limited EOR
investigations in the Queen, renewed developmental interest, and the success of neighboring fields’
EOR operations demonstrate the potential for increased and more sustainable oil recovery through the
use of CO2 foam. CO2 EOR in the Permian Basin has occurred throughout the past 40 years but lower
than expected oil recoveries are often reported due to gravity segregation, viscous fingering, and poor
sweep efficiency. The unfavorable CO2 properties, mostly density and viscosity, are main contributors
to these challenges which can be combated with foam for mobility control. Mobility control aims to
subdue the density and viscosity differences between CO2 and reservoir fluids through the use of
thickeners, gels, and/or foams.

The heterogeneous reservoir at Queen Field presents a substantial opportunity to demonstrate the
effectiveness of CO2 foam for mobility control. At the laboratory scale, CO2 foam has been shown to
overcome the unfavorable mobility ratio of supercritical CO2 in systems of variable heterogeneity
(Heller, 1966; Habermann, 1960; Bennion and Bachu, 2005, Fernø et al. 2014, 2015; AlMaqbali et al.
2015). Increased performance of CO2 foam for mobility control has also been demonstrated in more
heterogeneous systems, which included a fracture (Haugen et al., 2013; Figure 1). Despite the
favorable results of laboratory experiments, displacement mechanisms across larger scales are still not
well understood.

Numerous field pilot tests have been conducted utilizing CO2 foams with variable amounts of success
(Chou et al. 1992; Harpole et al. 1994; Stephenson and Graham, 1993; Hoefner and Evans, 1995;
Mukherjee et al. 2014). Most notably, difficulty has been encountered when monitoring the
propagation of CO2 foam in the reservoir and attributing additional oil recovery specifically to the
foam rather than to the increased volume of CO2 injected. The wide range of reservoir heterogeneities,
operational challenges, and complex fluid systems establishes the need for a more integrated
methodology for advancing CO2 foam technology. Integrating traditional laboratory techniques,
detailed core scale CO2 foam injection studies, geologic description and modeling, and validated
reservoir scale simulation models can provide new insights into the behavior of dynamic fluids across
multiple scales.

This study presents the initial work associated with a CO2 foam field pilot in the Permian Basin of
west Texas. This work emphasizes the utilization of substandard data in the form of old well logs,
sporadic production rates and field data, and relatively limited past reservoir characterization to
efficiently establish a baseline survey for the feasibility of CO2 foam injection. The application of a
reservoir management workflow, initial field pilot design, generation of static geologic and dynamic
reservoir models, and laboratory investigations comprise the first phase of a multi-phase project
aiming to advance the technology of CO2 foam for mobility control and CO2 storage.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Figure 1 Development of average oil saturation and differential pressure during waterflood and CO2
foam injection in a heterogeneous system (Haugen et al. 2013). Blue graph shows So as a function of
pore volume injected while red graph shows So vs pore volume injected during CO2 foam injection.
Differential pressure is plotted on the secondary y-axis.

Field Pilot Objectives and Expected Outcomes


The primary objective of the CO2 foam field pilot is to advance the technology of CO2 foam for
mobility control for EOR and CO2 storage. This includes demonstrating field wide viability and
establishing an integrated, multi-disciplinary workflow to guide decision making throughout the life
of the project which can be applied to similar field pilot tests.

Key components include:

 Integrating traditional laboratory techniques, detailed core scale CO2 foam injection studies,
improved reservoir characterization, and validated reservoir scale simulation models.
 Creating an improved model of CO2 foam processes through upscaling results from laboratory
scale to reservoir scale to assist in design, interpretation, and prediction of the field pilot.
 Developing and characterizing CO2 foam field systems for CO2 flooding of mature,
heterogeneous sandstone reservoirs aiming to increase sweep efficiency and CO2 mobility
control in interwell regions.
 Monitoring and validating the presence and propagation of foam in the reservoir and
attributing fluid displacement (i.e. increased oil recovery) to foam.
 Establishing cost efficient models for field wide implementation of CO2 foam EOR.

Application of a Reservoir Management Workflow


A reservoir management workflow has been applied to guide a systematic approach from data
gathering, model generation, updates, and decision making to final implementation and analysis of the
CO2 foam field pilot (Figure 2). The workflow considers multiple stages of project development and
establishes a generalized work plan for the field pilot research program. The current work considers
the reservoir characterization, generation of base static and dynamic models, and uncertainty analysis
which are all critical to initial field pilot design and establish a baseline survey for the feasibility of
CO2 foam for mobility control.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Figure 2 Reservoir management workflow used to guide project development, design, and
implementation. The current work considers the first three stages of the workflow which include
reservoir characterization, base model generation, and uncertainty analysis.

The workflow consists of several integrated components. First, review of historical field data and
consideration of field pilot well selection criteria permits a focused reservoir characterization study
for the pilot area. All available geologic and reservoir data are integrated to characterize the reservoir
in the pilot location. Next, the geologic framework is defined which guides the generation of base
geologic and reservoir models. Fluid studies are then initialized and integrated with the base models,
RCA, SCAL, PTA, and production data. Once base models are generated, they are calibrated by
utilizing a measure of heterogeneity to obtain representative realizations. Next, uncertainty analysis is
conducted using a probabilistic framework. This reduces and accounts for geologic and reservoir
uncertainty during performance forecasting.

Risks and opportunities are then considered and applied to decision making for pilot design and
execution. The pilot execution stage permits data acquisition which is used to update the pilot design
and understanding of risks and opportunities. Data gathered from pilot execution includes reservoir
pressure profiles, rates, monitor logging, and time lapse seismic. Analysis of this data will calibrate
models and validate the upscaling efforts to understand displacement mechanisms across multiple
scales.

The following sections present the first three stages of project of development. Stages are comprised
of the evaluation of field history, determination of field pilot selection criteria, reservoir
characterization, geologic and reservoir modeling, base model uncertainty analysis, flow simulation
initialization, and laboratory investigations. Field pilot design integrates each component, establishing
a multidisciplinary approach to project development.

Field History
Queen Field was discovered in the late 1940s and is located on the margin of the Central Basin
Platform in Permian Basin of west Texas (Figure 3). Cumulative oil production from the Queen

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
reservoir is over 2.5MMbo. The field covers approximately 3,100 acres and is considered to be
underdeveloped as the field is yet to be subjected to secondary recovery techniques. Pressure in the
reservoir was initially 1500 psia (hydrostatic) and reservoir behavior was typical of solution gas drive
with increasing gas-oil ratios and declining pressure. The rapidly declining reservoir pressure resulted
in lower than expected recoveries, slowing further operational development. However, field
operations progressed in the early 1990s as a waterflood field pilot was initiated. Infill drilling
established two 40 acre 5-spots, in the central parts of the field, consisting of six injectors and two
central producers. The project was unsuccessful as fill-up was never achieved, likely due to the halt of
operations due to project economics in the late 1990s. The production wells from the waterflood pilot
were subsequently shut in while other wells in the field maintained normal production status.

Figure 3 Location of Queen Field in the Permian Basin of west Texas. Red shaded area denotes field
location on the margin of Central Basin Platform in Pecos County, TX.

Currently, the field contains 74 wells and the reservoir has been depressurized to 350 psi. The shut in
producers of the waterflood pilot have been reopened, worked over, and are now online. The original
two water injectors of the waterflood pilot are back open and injecting produced water. Development
in the field is ongoing as the operator plans to implement a secondary waterflood in the western and
central parts of the field. A 1:1 primary to secondary recovery ratio is expected. Ten wells will be
converted to water injectors (sixteen in total) and four new wells will be drilled (establishing nine 40
acre 5-spots) by the end of Phase 1 of the operator’s current development plans. The waterflood will
aim to repressurize the reservoir, up to 1200 psi, and provide insight into injectivity and reservoir
response prior to CO2 foam injection.

Field Pilot Selection Criteria


Selection of a field pilot area initially begins with collaboration with field operators. Adhering to the
field development plan, it was deemed appropriate to analyze the central portion of Queen Field for
the pilot area location (Figure 4). This part of the field is undergoing the most rapid development as
four new wells will be drilled and the entire area will undergo secondary waterflood prior to CO2
foam injection. Drill core, modern petrophysical logs, fluid samples and analysis, RCA and SCAL
data will be acquired from at least one well in the pilot location area.

Criteria must be followed for the selection of successful pilot wells within the designated field area.
The selected wells must contain appreciable reservoir characteristics and show continuous distribution
of reservoir flow zones between wells. Field pilot wells should also be in close proximity and
demonstrate connectivity. An examination of multiple patterns in the field included the analysis of
historical production data, well logs, core and thin section data, and correlation of flow zones. An
inverted 40 acre 5-spot pattern has been identified with well QFS-1 as the injector and wells QFS-2, 3,
4, and 5 as the producers (Figure 4).

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Figure 4 Central wells at the end of Phase 1 of development. Dashed line shows selected field pilot
pattern consisting of one central injector and four surrounding producers.

QFS wells 1, 2, and 3 showed favorable distribution of reservoir flow zones and appreciable
connectivity (Figure 5). QFS production wells 3 and 4 were drilled in September 2016 and completed
in each reservoir zone. Modern logs, routine core analysis, and fluid samples were collected and are
being analyzed. The use of new wells is advantageous, as it not only allows data collection but
provides insights into current fluid saturations, reservoir pressure, and essential information to
advance the reservoir characterization and update base geologic models. A detailed integration of
newly acquired field data will follow in the second phase of project development. A review of recent
historical production data permits the following discussion. Production profiles of the central portion
of Queen Field, including pilot wells and peripheral wells, are shown in Figure 6.

QFS-1 was drilled in March 1984 as a production well and completed to a depth of 3700ft. The well
initially produced 600 bbls/month (Figure 7). However, production fell rapidly over the next three
years and the GOR increased before production rates steadied to approximately 100bbls/month until
June 1994. During steady oil production the GOR increased, reaching a maximum value of 5.6
Mcf/month.

The well was converted to water injector, developed on a 40 acre 5-spot, in preparation for the pilot
waterflood in July 1994. An average of 5000 bbls/month of water was injected at a pressure of 1400
psig. During injection, surrounding production wells showed favorable response as wells produced
above a 1:1 ratio of primary to secondary recovery. However, reservoir pressure rates were not
regularly recorded providing little insight into reservoir response. The overall poor quality and
sporadic nature of the available data for the pilot waterflood does not allow for determination of a
baseline survey for injectivity.

QFS-2 was drilled as a production well in July 1997 and completed to depth of 3700ft. The well was
developed on 40-acre spacing within the waterflood pilot 5-spot. Large discrepancies are seen in
reported production data as they’ve been reported from multiple operators to the Texas Railroad
Commission. However, comparison with neighboring wells and discussion with the current operator
permits the following discussion.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Figure 5 Cross section of three wells in the selected field pilot 5-spot pattern with gamma ray,
neutron, and density porosity logs showing correlation of reservoir zones. Colored zones indicate
individual reservoir flow zones while light shaded areas represent non reservoir intervals.

100000 100000
Start of Pilot WF
Liquid Production (bbl/mon)

Gas Production (Mcf/mon)

10000 10000

1000 1000

100 100

Oil Prod
10 10
Water Prod
Mar-77 Mar-87 Mar-97 Mar-07 Mar-17
Gas Prod
Dates

Figure 6a Oil, water, and gas production for the central portion of Queen Field. Red dashed line
denotes the start of the pilot waterflood in July 1994.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
18 50000
16
Start of Pilot WF 45000
40000

Water Injection (bbl/mon)


14
35000
12
GOR (Mcf/bbl)

30000
10
25000
8
20000
6
15000
4 10000
2 5000
0 0 GOR
Mar-77 Mar-87 Mar-97 Mar-07
Dates Water
Injection

Figure 6b GOR and injection profile for the central portion of Queen Field. Red dashed line denotes
the start of the pilot waterflood in July 1994.

700 10000
9000
600

Water Injection (bbl/month)


8000
Oil Production (bbl/mon)

500 7000
6000
400
5000
300
4000

200 3000
2000
100
1000
0 0
Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94 Mar-96 Mar-98 Oil Prod

Dates Water Injection

Figure 7a Well QFS-1 historical oil production from March 1984 to May 1994 and water injection
from July 1994 to January 1997.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
8

6
GOR (Mcf/bbl)

4 GOR

0
Mar-84 Mar-86 Mar-88 Mar-90 Mar-92 Mar-94
Dates
Figure 7b Well QFS-1 historical producing GOR from March 1984 to June 1994.

Monthly production began at 150 bbls and fluctuated over the next 6 months, reaching a maximum of
300 bbls. Production rapidly declined in March 1998 producing an average of 10 bbls/month until
December 2002. GOR ratios have fluctuated throughout production but remained between 1-8
Mcf/bbl, reaching a maximum of 17 Mcf/bbl in September 1997 (Figure 8).

QFS-3 was drilled in June 2008 and completed to a depth of 3642ft. The well initially started
producing 300 bbls but oil production rapidly declined to an average of 100 bbls/month until
December 2010. During this time, GOR were between 2-3.5 MCF/bbl. No production is reported
again until February 2013 when the well produced 24 bbls at a GOR of 3 Mcf/bbl. Production steadily
declined to approximately 1bbl/month by the end of 2013 (Figure 9).

Reservoir Characterization
The evaluation of historical field data, selected pilot pattern location, and integration of available
geologic and reservoir data permitted a detailed reservoir characterization to be established for the
field area of the pilot test. The Queen Formation is located at depths of 3100ft to 3500ft. The net pay
consists of approximately 120ft of tidally deposited fine grained sandstone interbedded with dolomite
and anhydrite. Small fluctuations in sea level result in the layering of the reservoir with distinct
siliciclastic flow zones separated by low permeability and low porosity carbonate and evaporites
deposits. Fluid flow within the reservoir can be described through the identification of flow zones,
based upon layers with appreciable reservoir characteristics within the Queen. Average reservoir and
fluid properties are found in Table 1.

Four flow zones constitute the main pay and are defined as follows:
 Porosity cut-off ≥ 11%
 Gamma ray ≥ 45 API Units
 Average thickness = 20ft
 Matrix density = 2.66g/cm3 – 2.70g/cm3

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
350 20

18
300
16
Oil Production (bbl/mon)

250 14

GOR (Mcf/bbl)
12
200
10
150
8

100 6

4
50
2

0 0
Jun-97 Jun-98 Jun-99 Jun-00 Jun-01 Jun-02 Oil Prod
GOR
Dates

Figure 8 Well QFS-2 historical oil production and producing GOR.

400 30

350
25
Oil Production (bbl/mon)

300
20
GOR (Mcf/bbl)

250

200 15

150
10
100
5
50

0 0
Jun-08 Oct-09 Mar-11 Jul-12 Dec-13 Oil Prod
Dates GOR

Figure 9 Well QFS-3 historical oil production and producing GOR.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Table 1 Reservoir characteristics of the Queen Formation

Reservoir Characteristic Value


Depth 3300 ft
1 – 30 mD
Permeability
Ave. 16 mD
3 – 35 %
Porosity
Ave. 20 %
Pay Thickness 120 ft
Reservoir Pressure 350 psia
Temperature 95oF
Oil Gravity 36o API
Initial Oil Saturation 0.56
Initial Water Saturation 0.44
Oil viscosity (reservoir condition) 2.50 cP
Formation Brine Salinity 40,000 ppm

Geologic Framework and Depositional Setting


The geologic framework was established through the analysis of all available geologic and reservoir
data to better understand the distribution of the Queen sandstone intervals throughout the field area.
The field lies on the southern margin of the Central Basin Platform which extends into the Sheffield
Channel of southwest Texas (Figure 3). The platform is a northwest-southeast trending, doubly
plunging anticline with structural and stratigraphic traps acting to restrict and cap fluid migration.

The Queen Formation is Guadalupian (Artesia Group) in age and predominantly consists of fine to
medium grained sand units, interbedded with variable amounts of carbonates and evaporites. The
Queen is overlain by carbonates of the Seven Rivers Formation and underlain by the, predominately
carbonate, Grayburg Formation (Figure 10). Distinct layers of higher porosity fine grained sandstones
and lower porosity carbonates and evaporites are found throughout the field. Sediments were
deposited shelfward of the Guadalupian reef complex in depositional settings dominated by
carbonates in outer shelf positions, evaporites and siliciclastics in middle shelf positions, and
siliciclastics in the most shelfward positions. The wide occurrence of siliciclastic deposits, over other
Guadalupian formations, provides the basis for interpreting the Queen as recording a period of sea
level lowstand (Nance, 2004).

Figure 10 Stratigraphic correlation chart for the Artesia Group in the Permian Basin

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Reservoir Modeling
The objective of reservoir modeling is to describe the spatial distribution of parameters through the
integration of wellbore data with the geologic model. When available, conventional core and thin
section data were calibrated to available logs (neutron, density, and resistivity). Correlation of
reservoir zones throughout the pilot area provided the basis for reservoir geometry and structure
(Figure 5).

The Queen is readily identifiable on gamma ray logs by upper and lower bounding carbonate and
anhydrite formations which show low gamma ray response in comparison to a high gamma ray
response by the siliciclastic facies (Figure 11). High amounts of K-feldspar in the sandstone results in
the high gamma response which are easily traced and correlated throughout the study area. The
tabular nature of the sandstone intervals corroborates their correlation throughout the pilot area.

The repetitive cyclic nature of the Queen Formation permitted a simple stratigraphic layering
approach to be applied to describe the structure and distribution of reservoir zones. Petrophysical well
logs from 60 wells and conventional core analysis and thin sections from one well (QFS-5) provided
the bulk of the data for classifying each layer, identifying cycle boundaries, and conditioning
petrophysical properties to facies (Figure 11). 14 cycles were identified which are composed of basal
fine to medium grained sandstones with predominately carbonate facies occurring at the tops. Cycles
range in thickness from 10-25ft and occur at critical scales where facies correlations can be made and
petrophysical properties can be modeled.

Figure 11 Gamma ray, density, and neutron porosity log for one well in the selected pilot pattern.
Blue lines indicate the top of depositional cycles which are characterized by basal fine grained
sandstone and topped with carbonate facies. Note high gamma responses from sandstone intervals in
comparison to carbonate and evaporate zones.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
A base model was established using the geocellular modeling capabilities of Petrel E&P Platform
(Schlumberger, 2015.1). Several stochastic realizations were generated for each layer using flow
zones and identified cycles as stratigraphic constraints. The tops of each flow zone were mapped and
populated with calculated porosities and permeabilities to honor the geologic framework and populate
the grid in interwell regions. The use of stratigraphic constraints in stochastic simulation allows
interwell regions to be more accurately modeled through the integration of systematically stacked
rock units within the geologic framework.

The base grid has dimensions of 72 x 72 x 21 and contains approximately 109,000 grid cells. Grid cell
sizes are 50ft x 50ft and each reservoir layer is continuous. Thicknesses of layers in the base model
are dependent upon stratigraphic units which range in thickness from 5-35ft with an average of 20ft in
the Z-direction. Base models were calibrated and screened to reduce geologic uncertainty in reservoir
performance prediction.

Calibration and Screening of Static Models


A limited amount of historical data is available to calibrate the reservoir model for the pilot area.
Therefore, we use a geostatistical simulation method to create many equally probable realizations,
while preserving the first moment (mean) and second moment (variance) of available well data. This
spatial distribution of static properties in the model governs the flow pattern for each realization. The
limitation to this approach is that the time required to calibrate all realizations to historical data and
run production forecasts increases significantly with the number of realizations in the ensemble, and
can be prohibitive. For this reason, the realizations are ranked based on a heterogeneity measure, and
retain a few representative realizations for pilot simulation study.

Heterogeneity measures can be classified into two broad categories - static and dynamic. Static
measure is based on data derived from core or well log. Dynamic measure, on the other hand, uses a
flow experiment (physical or numerical simulation) to capture the distribution of flow paths directly.
We use streamline simulation and time-of-flight information to generate flow capacity – storage
capacity curve or Lorenz plot for each realization. A dynamic Lorenz coefficient (Schmalz and
Rahme, 1950; Lake, 1989) is calculated, which ranges between 0 for homogeneous reservoir to 1 for
infinitely heterogeneous reservoirs. So, rather than working with a single scenario, we use this
measure of heterogeneity to retain a few representative realizations, combined with screening used
against dynamic data to obtain robust prediction forecasts.

Flow Simulation Initialization


It is beyond the scope of this paper to discuss in detail the results of the first stages of flow
simulations. Rather, the following sections present the initialization of establishing flow models and a
workflow for ranking static model realizations which are tuned with static and dynamic heterogeneity
measures, preparing them for flow simulations to match and predict reservoir performance.

Probabilistic workflow
It is natural for an integrated reservoir model to have uncertainty in performance prediction owing to
availability of finite amount of data for reservoir characterization, and use of a less-detailed model to
capture the flow dynamics as a trade-off between complexity and simulation runtime. Given these
limitations, the probabilistic modeling framework used in this work and presented here aids in
quantifying the impact of uncertainties on the expected performance prediction.

Upon completion of the dynamic base model, the first step is to identify the sources of uncertainty.
The members of a multi-disciplinary team agree upon the range and probability density distributions
for the uncertainty parameters. A sensitivity analysis is then performed by varying one-variable-at-a-
time from the base value, to get insights into relevance of individual uncertainty parameter for further
scoping studies. This is used to rank the uncertainty parameters to decide upon the significant
uncertainty sources that should be carried forward to uncertainty quantification stage.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
We apply the Latin Hypercube method for sampling uncertainty space and simulations were run to
quantify the impact of the uncertainty on key performance indices. This approach flags relevant risks
and opportunities, and proves valuable in steering decisions towards optimum pilot design and
operation.

Fluid Model Initialization, MMP


Separator fluid samples were collected in September 2016 for fluid characterization studies.
Compositional analyses were performed on the atmospheric separator oil and separator gases using
gas chromatography methods (Figure 12). The molecular weight and density of the flashed liquid
phase were measured.

Further laboratory experiments were conducted on a recombined sample, made of separator gas and
dead oil sample. The recombination GOR was arrived at using measured fluid composition and
default parameters for EoS model. A certain volume of recombination fluid was flashed to standard
condition, to ensure that compositions of the flashed gas and flashed oil match the composition for
samples collected at surface. After recombination validation, constant composition expansion
experiment was performed at the reservoir temperature. Saturation pressure was determined visually
and graphically from the CCE experiment.

MMP prediction by multiple mixing cell method and method of characteristics suggest that first
contact miscibility cannot be achieved until the reservoir pressure is increased significantly; however,
there is a possibility of achieving miscibility through multiple contacts. The generated fluid model
will be used in flow simulations to gain insights into the performance of CO2 foam at reservoir scale
and will be discussed in second phase of project development.

100
90 Sep. Gas Sample Sep. Oil Sample

80
Composition, mol %

70
60
50
40
30
20
10
0
N2 CO2 H2S C1 C2 C3 i-C4 n-C4 i-C5 n-C5 C6 C7+
Components

Figure 12 Well QFS-5 atmospheric separator oil and gas sample compositional analysis used in fluid
characterization studies.

Laboratory Investigations
Laboratory work is being performed to test CO2 foam systems for mobility control to optimize EOR
potential as a part of the ongoing reservoir management workflow. Primary depletion of the Queen
has resulted in a depressurized reservoir from initially 1500 psi (~103bar) to a current state of 350 psi
(~24bar). The ongoing waterflood is anticipated to repressurize the field back to original hydrostatic
pressure at 1200 psi (~82bar). Associated laboratory work therefore aims to determine effects of
reservoir pressure on CO2 foam performance for field implementation.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Due to anticipated variable reservoir pressures, this first set of experimental work aims to determine
the effect of foam on oil recovery at two different pressures. Available data that determine current
reservoir pressure is sparse, and it is uncertain how reservoir pressure will respond to waterflood.
Laboratory CO2 foam injections were therefore conducted at 600 psi (~41bar) and at near original
hydrostatic pressure (1200 psi/~82bar) to determine foam behavior at repressurized conditions and to
investigate the potential for miscible flooding.

Core and Fluid Properties


Corefloods were conducted on 1.5” diameter homogeneous outcrop Bentheimer Sandstone cores.
Porosity value of the Bentheimer is on average 23%, equal to that of the Queen Sandstone, but with
variable permeability. Reservoir crude oil was used and synthetic Permian brine analogous to that of
the field. An anionic surfactant Petrostep C1 was chosen as foaming agent and a solution was made
by adding 1 wt% of surfactant to Permian brine. CO2 gas was received at purity of 99,999%.

The sandstone core plugs were washed, dried and saturated with brine under vacuum. Porosity was
calculated by measuring the weight of the cores before and after saturation provided the density of the
brine. Absolute permeability was determined by injecting brine and measuring the differential
pressure across the core. Core names and properties are found in Table 2.

Table 2 Core identification and properties from routine core analysis.


Core ID Porosity [ф] Permeability [D] Length [cm] PV
FT_S1 0.23 ±1.18E-03 1.9 ±0.1 15.2 ± 2.0E-05 39.3 ±0.2
FT_S2 0.23 ±1.18E-03 2.2 ±0.1 15.3 ± 2.0E-05 39.4 ±0.2
FT_S3 0.23 ±1.16E-03 2.3 ±0.1 15.3 ± 2.0E-05 38.6 ±0.2
Instrumental uncertainty given as ±
Experimental Details and Procedures
CO2-foam injections were performed at 35°C in a high pressure/high temperature setup inside a heat
cabinet. The cores were wrapped in aluminum to reduce CO2 damage on the core holder rubber sleeve
applying a radial overburden pressure of 20bar above pore-pressure at all times. A backpressure
regulator was used to keep a constant system pressure of either 600 psi or 1200 psi (41bar or 82bar).
Differential pressure was measured during all injections by absolute pressure transducers positioned at
the inlet and outlet of the core.

Oil drainage was performed with reservoir crude oil. Five pore volumes (PVs) of oil were injected
until an irreducible water saturation of ~0.25 was achieved. Secondary waterflooding (1PV) was then
performed with a superficial velocity of 4 ft/day (~56ml/h). Saturation data and experimental
conditions are listed in Table 3. Tertiary CO2 foam injection was performed co-injecting surfactant
solution and CO2 gas at a foam quality of 80%. The total rate during co-injection was kept equal to
waterflooding (i.e. 45ml/h CO2 and 11ml/h surfactant solution). A foam generator of packed
unconsolidated sand grains (diameter of 250-500µm) was used to pre-generate foam prior to entering
the core.

Enhanced Oil Recovery and Mobility Reduction by CO2 Foam


The effect of pore pressure was studied in two tertiary CO2 foam injections focusing on the reduction
of gas mobility by foam, CO2-oil miscibility, and oil recovery (Figure 13). The waterflood recoveries
were approximately ~50%OOIP after injecting 1 PV of brine.

At a pore-pressure of 41bar, CO2 foam produced an additional 3.8%OOIP after a total of 1.1 PV
injected (open blue circles, Figure 13). The low incremental recovery and the amount of PVs injected
suggest poor sweep efficiency by the generated foam. At 41bar the injected CO2 is in gas phase and
the density and viscosity is low which results in limited mobility reduction and a low apparent foam
viscosity (Table 3). With a pore-pressure of 82bar, the oil production during CO2 foam was
accelerated compared with 41bar, whereas the incremental recovery was similar. An additional

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
4.1%OOIP was recovered after injecting 0.31 PVs (open red circles, Figure 13). The advanced
response in oil recovery indicates a more favourable mobility ratio between the injected CO2 and
reservoir fluids. The density and viscosity of the CO2 gas increases as it transitions into supercritical
state at 35°C and 82bar (National Institute of Standards and Technology, 2016). The higher density
and viscosity of the CO2 combined with the foaming agent reduce gas mobility significantly and
improve sweep efficiency.

Table 3 Saturation data, experimental conditions and results. All corefloods and foam quality scans
were conducted at a constant temperature of 35oC
Core Rf,tot Rf,CO2 foam µapp,WF µapp,CO2foam MRF P µCO2,gas
Swi So,wi So,WF So,CO2foam
ID [%OOIP] [%OOIP] [Pa*s] [Pa*s] [Pa*s] [bar] [Pa*s]4)

1.63*E-
FT_S11) - - - - - - - - - 82
05
0.26 0.74 0.38 0.35 0.0041 3.93*E-
52.9 3.8 0.0004 17.6
FT_S22) ±1.39E- ±1.39E- ±3.40E- ±3.78E- ±2.25E- 41
03 03 03 03
±0.5pp ±0.5pp
04
±1.96E-03 ±357.9 0-5
0.24 0.76 0.37 0.34 55.7 4.1 0.0042 0.0041 113.2 3.93*E-
FT_S33) ±1.28E- ±1.28E- ±3.43E- ±3.82E- ±2.50E- 82
03 03 03 03
±0.6pp ±0.6pp
04
±2.65E-02 ±655.3 05
1)
Foam stability w/ supercritical CO2 2) EOR w/CO2 gas 3) EOR w/ supercritical CO2 4) NIST, 2016 Instrumental uncertainty given as ±

Figure 13 Dynamic recovery profiles during secondary waterflooding and tertiary CO2 foam
injection at 41bar (blue graph) and 82bar (red graph) as a function of pore volumes injected. The
secondary y-axis shows the differential pressures across the cores during injection. Temperature is
constant at 35°C.

Endpoint mobility reduction factor (MRF) was calculated based on apparent foam viscosity at the end
of CO2 foam injection when the pressure differential across the core reached steady state. The values
are listed in Table 3. At 41bar, CO2 gas mobility was reduced by a factor of 17.6, which means that
the apparent viscosity of the foam was approximately 17 times greater than that of CO2 gas at same
conditions. Correspondingly, at 82bar, the MRF was 113, five times greater than the foam generated
at 41bar. The reduction in gas mobility also reduced the need for CO2 and surfactant to be injected to

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
improve oil recovery, as observed in Figure 13. Repressurizing the Queen reservoir will therefore be
favorable for oil recovery by CO2 foam considering the economic limitations of CO2 and surfactant
usage.

The low differential pressures observed during co-injection indicate the generation of weak foam
(Figure 13). During CO2 foam flooding the transition from weak to strong foam is generally detected
through significant increases in pressure drops across the cores (Sheng, 2013). The low pressure
response by CO2 foam results from its apparent viscosity being less than or equal to the viscosity of
brine at the end of waterflooding (Table 3). A stability experiment was therefore performed at 82bar
and 35°C to isolate the parameters that may affect the behavior and generation of CO2 foam.

Foam Quality
A core was saturated 100% with surfactant solution (i.e. no was oil present), and CO2 foam was
generated during co-injection of CO2 and surfactant solution using foam qualities from high-to-low
gas fractions (Figure 14). The MRF at a foam quality of 80% was 227, approximately double the
observed MRF during the co-injection for EOR at 82 bar (red circle in Figure 14).

Figure 14 Mobility reduction factor (MRF) as a function of foam quality at 82bar and 35°C. The
green graph shows average MRF as foam quality is altered from high-to-low gas fractions. The red
point at 80% foam quality represent the MRF calculated from the equivalent EOR experiment at
82bar.

The detrimental effect of reservoir crude oil on foam is well documented in the literature, and is the
likely explanation for the reduced MRF observed during CO2 foam EOR test (Figure 13) compared
with foam stability (Figure 14). If foam is degraded, surfactant may partition into the oil or create
emulsions between the water and oil phase present (Nikolov et al. 1986). Sweep efficiency by foam is
reduced and oil recovery is low. Extensive surfactant screening is ongoing to determine optimal
surfactant for following EOR experiments on reservoir core material.

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
MMP predictions indicated that first contact miscibility could not be reached until reservoir pressure
was significantly improved. Multiple contact miscibility could, however, be achieved if the reservoir
was successfully repressurized to its initial hydrostatic conditions as shown by the multiple mixing
cell method. Nevertheless, the preceding core-scale experiments suggest that the fluid system will
remain immiscible at both current and anticipated reservoir pressures. Repressurizing the field may
not necessarily improve the miscibility between the CO2 and reservoir crude oil, but it will increase
the sweep efficiency of the CO2 foam improving reservoir performance.

Conclusions
A CO2 foam field pilot research program has been initialized to test and advance the use of CO2 foam
for mobility control for EOR. This works presents the first phase of project development and provides
valuable insight into the design and implementation of a CO2 foam field pilot in an underdeveloped,
mature oil field for which substandard data is currently available.

An improved reservoir characterization, analysis of historical well performance, and collaboration


with the operators result in the selection of an inverted 5-spot pattern for piloting. The pilot well
selection criteria considered that the wells show appreciable reservoir characteristics, continuous
distribution of reservoir flow zones, and close proximity. Further, the inclusion of two newly drilled
wells provides insight into current field fluid saturations, reservoir pressure, and essential information
to advance the reservoir characterization and update base geologic models. Geologic and reservoir
models were generated and are being calibrated with heterogeneity measures to determine
representative realizations. A probabilistic workflow will be employed to flag relevant risks and
opportunities and aid in steering decisions towards optimum pilot design and operation.

Laboratory work investigating foam’s behavior at variable pressures found that repressurizing the
reservoir will result in more favorable CO2 foam behavior as it will recover oil more effectively. This
will be viable considering the economic limits on CO2 and surfactant usage, though it may not
necessarily improve the miscibility between the injected CO2 and reservoir crude oil. Future
experimental work will utilize heterogeneous field core, reservoir crude oil, and brine to analyze foam
performance and provide insights into injection strategies and foam model input parameters.

Acknowledgements
The authors wish to acknowledge the Norwegian Research Council CLIMIT program for financial
support under grant number 249742 - CO2 Storage from Lab to On-Shore Field Pilots Using CO2-
Foam for Mobility Control in CCUS and industry partners; Shell E&P, TOTAL E&P, and Statoil. The
authors also thank the field operator and Schlumberger for technical support.

Nomenclature
Swi Irreducible water saturation
So,wi Oil saturation at irreducible water saturation
So,WF Oil saturation after waterflooding
So,CO2foam Oil saturation after CO2 foam injection
Rf,tot Total oil recovery
Rf,CO2 foam Incremental oil recovery by CO2 foam
µapp,WF Apparent viscosity of water at the end of waterflooding
µapp,CO2foam Apparent viscosity of CO2 foam at the end of co-injection
µCO2,gas Apparent viscosity of CO2 gas

Abbreviations
EOR Enhanced Oil Recovery
SCAL Special Core Analysis
RCA Routine Core Analysis
PVT Pressure Volume Temperature
EoS Equation of State
MMP Minimum Miscibility Pressure

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
CCE Constant Composition Expansion
PTA Pressure Transient Analysis
GOR Gas-Oil Ratio
MRF Mobility Reduction Factor
PV Pore Volume
OOIP Original Oil in Place
NIST National Institute of Standards and Technology

References
AlMaqbali, A., Agada, S., Geiger, S., Haugen, Å., and Fernø, M. A. [2015] Modelling Foam
Displacement in Fractured Carbonate Reservoirs. SPE International Petroleum Conference Abu
Dhabi, UAE 9-12 September 2015. DOI:10.2118/177600-MS

Bennion, B. and Bachu, S. [2005] Relative Permeability Characteristics for Supercritical CO2
Displacing Water in a Variety of Potential Sequestration Zones in the Western Canada Sedimentary
Basin. SPE Annual Technical Conference and Exhibition, 9-12 October 2005, Dallas, Texas. SPE-
95547

Chou, S., Vasicek, S., Pisio, D., Jasek, D., and Goodgame, J. A. [1992] CO2 Foam Field Trial at North
Ward-Estes. SPE Annual Technical Conference and Exhibition, 4-7 October 1992, Washington, D.C.
SPE-24643-MS

Fernø, M. A., Gauteplass, J., Pancharoen, M., Haugen, Å., Graue, A., Kovscek, A. R., & Hirasaki, G.
[2014] Experimental Study of Foam Generation, Sweep Efficiency, and Flow in a Fracture Network.
Society of Petroleum Engineers Journal 21 (04). DOI: 10.2118/170840-PA

Fernø, M.A., Steinsbø, M., Eide, Ø., Ahmed, A., Ahmed, K., Graue, A. [2015] Parametric Study of
Oil Recovery during CO2 injections in Fractured Chalk: Influence of fracture permeability, diffusion
length and water saturation, Journal of Natural Gas Science & Engineering (2015), DOI:
10.1016/j.jngse.2015.09.052.

Habermann, B. [1960] The Efficiency of Miscible Displacement as a Function of Mobility Ratio.


AIME 1960, 219, 264-272

Harpole, K.J., Siemers, W.T., and Gerard, M.G. [1994] CO2 Foam Field Verification Pilot Test at
EVGSAU: Phase IIIC- Reservoir Characterization and Response to Foam Injection. SPE/DOE 9th
Symposium on Improved Oil Recovery, 17-20 April, Tulsa, Oklahoma, USA. SPE/DOE-27798

Haugen A., Mani, N., Svenningsen, S., Brattekås, B., Graue, A., Ersland, G., and Fernø, M.A . [2014]
Miscible and Immiscible Foam Injection for Mobility Control and EOR in Fractured Oil-Wet
Carbonate Rocks. Transport in Porous Media 2014 104, 109-131. DOI: 10.1007/s11242-014-
0323-6

Heller, J.P. [1966] Onset of Instability Patterns Between Miscible Fluids in Porous Media. Journal of
Applied Physics 1966, 37, 1566-1579.

Heller, J.P., Boone, D.A., and Watts, R.J. [1985] Testing CO2-Foam for Mobility Control at Rock
Creek. SPE 1985 Eastern Regional Meeting, 6-8 November, Morgantown, West Virginia, USA, SPE-
14519

Hoefner, M.L. and Evans, E.M. [1995] CO2 Foam: Results from Four Developmental Field Trials.
SPE Reservoir Engineering, November 1995, 273-281. SPE- 27787

Lake, L. [1989] Enhanced Oil Recovery, Prentice Hall

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway
Mukherjee, J., Norris, S.O., Nguyen, Q.P., Scherlin, J.M., Vanderwal, P.G., and Abbas, S. [2014] CO2
Foam Pilot in Salt Creek, Natrona County, WY: Phase I: Laboratory Work, Reservoir Simulation, and
Initial Design. SPE Improved Oil Recovery Symposium , 12-16 April, Tulsa, Oklahoma, USA. SPE-
169166-MS

Nance, H.S. [2004]. Guadalupian [Artesia Group] and Ochoan Shelf Succession of the Permian Basin:
Effects of Deposition, Diagenesis, and Structure of Reservoir Development. PhD Dissertation. The
University of Texas at Austin.

National Institute of Standards and Technology. [2016]. Thermophysical Properties of Fluid Systems.
Retrieved January 20, 2017, from NIST Chemistry WebBook:
http://webbook.nist.gov/chemistry/fluid/

Nikolov, A. D., Wasan, D. T., Huang, D. W., & Edwards, D. A. [1986]. The Effect of Oil on Foam
Stability: Mechanisms and Implications for Oil Displacement by Foam in Porous Media. 61st Annual
Technical Conference and Exhibition. New Orleans, LA, USA: Society of Petroleum Engineers.

Schmalz, J.P. and Rahme, H.D. [1950] The Variation in Waterflood Performance with Variation in
Permeability Profile. Production Monthly, 15[9], 9-12.

Sheng, J. J. [2013]. Enhanced Oil Recovery Field Case Studies. Tokyo: Elsevier.

Stephenson, D.J. and Graham, A.G. [1993] Mobility Control Experience in the Joffre Viking Miscible
CO2 Flood. SPE Reservoir Engineering, August 1993, 183-188. SPE- 23598

IOR 2017 – 19th European Symposium on Improved Oil Recovery


24-27 April 2017, Stavanger, Norway

You might also like